Hacker News Reader

Comment by _7acn | Hacker News Reader

Just under 30 years ago, when I was starting my IT studies, I had an older colleague who was a great authority to me. When I began learning about RDBMS options, I called him to ask, „What do you think about Oracle?”. He just shouted, „Total crap!!!” and that was enough for me. Since then, to this day, I’ve never touched Oracle.

View on HN

I don't think it's necessarily fair to say Oracle is total crap. For decades they were more performant and had more features than most anything else on the market. They supported scale-out clustering on linux with Oracle RAC. They were early adopters of high performance NFS moving the NFS stack out of the linux kernel and directly into the database with DirectNFS. They built a clustered filesystem for block-based clusters (OCFS).

HOWEVER, the way they run their business is horrible. Oracle the product was various versions of awesome to just OK. Oracle the business is a modern-day mafioso shakedown.

In the early 2000s, nearing the launch of The Sims Online, Oracle sent their sales team to Maxis. I was a newish employee and wasn't involved in the decision process, but I overhead much of the conversation (my desk was just outside Will Wright's office[1]). Every time the Oracle rep said "if you invest in RAC..." (emphasis theirs) I absolutely cringed at the blatant attempt at NLP.

They bought it, and it was a disaster. Turns out you can't just take a system designed without partitioning in mind and put it on a distributed database. Not that it mattered; the system saw a tiny fraction of its expected usage. The whole game was an absurd amount of money spent on nothing.

To this day I have never used another Oracle product.

[1] I don't think Will Wright was involved in that decision either; I think he was fully checked-out by then and in any case ops wasn't his bailiwick. The conversation just happened to take place in his office.

People either love or hate Oracle back then. I was in the love group. However, in recent days, Oracle is pretty irrelevant, imho.

I'd rather pay money to EnterpriseDB than give it to Redwood Shores.

It doesn't matter how great their database was. If they want to shake down businesses (which they are known to do), then don't have anything to do with them. Ever.

The type of business that rests on its laurels and believes ratcheting continuously-increasing rent seeking at the expense of goodwill will soon find themselves without new customers and possibly without a market because they took their self-assured dominance for granted. Hubris. Cfengine and Tripwire both found that out the hard way.

> They supported scale-out clustering on linux with Oracle RAC.

That never worked right. Amazon got burned trying to use it, Sberbank in Russia went down hard for several days because RAC just crashed, etc.

Yeah every argument about Oracle being bad is business related, not tech.

Mmm, one of the most memorable stories I've ever read on HN is about Oracle being bad at tech.

https://news.ycombinator.com/item?id=18442941

I remember that one!

Yeah, this is legendary comment

The one job where I got to experience an Oracle database in action, back around ought four (allegedly one of the good periods for Oracle's db tech):

A) Performance was so terrible by default the DBAs were cops and had given up on real RDBMS performance tools such as indexes and keys (primary and foreign) and instead used a ton of arcane low level file management tools like Oracle was some sort of "build your own DB kit" and the out of the box one was terrible so they needed a bespoke hand built one

B) I was once in a CLI running some sort of very basic Select statement (no joins, because again, no keys; single table; maybe a couple of columns) and watched in horror as an entire Java-based debugger IDE not installed on my extremely locked down work machine launched, spun up, and dived deep, deep into a terrifying stacktrace comprising an awful mix of C++ and Java code. To this day I don't know how or why a Select statement (that worked the second time) crashed so hard into Oracle's own source code. It is possibly related to part A above, but I'm still not sure. I don't know why the debugger symbols much less actual source code from Oracle were even available when it did crash that hard. I did know at the time that actually debugging Oracle's code was very far above my pay grade and in addition to sending the debugger IDE they should have also attached a large check.

I think of that experience often when people tell me that Oracle actually has good tech. If it weren't obvious from how awful their tools are to use as a user (a different previous employer used Oracle's expensive time/attendance/payroll tools and those were the clunkiest, worst web apps to use), that brief, weird horror story of the Select statement in a CMD.EXE window REPL bringing up the wildest stack trace in a debugger that didn't exist on my machine will always leave me feeling doubtful about that.

Oh no, they have shoveled some shit in their day for sure. I mean their ERP, their enterprise java bean stuff. But their flagship database product was always quite good, especially if you had staff or consultants who really knew it.

I probably should have said that most conversations… don’t reach the tech stage and stop at their business practices.

IME you don't usually choose Oracle. Someone choose it for you, and you can move somewhere else if you really hate it.

The SQL/PSM standard comes from Oracle, and is variously implemented in PostgreSQL, MariaDB, IBM Db2, and more.

https://en.m.wikipedia.org/wiki/SQL/PSM

"SQL/PSM is derived, seemingly directly, from Oracle's PL/SQL. Oracle developed PL/SQL and released it in 1991, basing the language on the US Department of Defense's Ada programming language. However, Oracle has maintained a distance from the standard in its documentation. IBM's SQL PL (used in DB2) and Mimer SQL's PSM were the first two products officially implementing SQL/PSM."

Reply on news.ycombinator.com